How to run effective industry-enriched project-based learning in STEM

If we are serious about diversifying science, technology, engineering and mathematics, we need to look beyond simply who gets admitted. We must pay closer attention to what happens once students arrive on campus. For many learners – especially those from non-traditional pathways, diverse cultures or adjacent disciplines – STEM can feel like a gated community: heavy on prerequisites, hidden norms and success measured by narrow academic performance. 

Industry-enriched project-based learning flips this. It treats real-world problems as core learning contexts, values collaboration and communication as essential technical skills, and creates space for diversity to enable opportunity.

But “just add industry projects” is not a plan. Industry projects need careful design. Without it, unmanaged partnerships can amplify inequity and create reputational risk. When one-to-one supervision doesn’t scale, it overloads staff. 

• Develop global citizens with virtual exchanges
• Build lasting partnerships with industry, from the ground up
• How can universities show the impact of research and boost connections with the public?

Our approach for scaling applied data science projects, informed by research and refined by experience, is applicable across STEM and higher education in general. We work with more than 100 students annually, with increasingly diverse cultures and backgrounds. Unlike typical industry placements, which often rely on ad hoc partnerships with variable expectations, this approach separates project modes, standardises core processes and distributes supervision at scale without sacrificing rigour or equity.

Designing a curriculum that is integrated with industry

The foundation lies in a curriculum that accommodates student diversity while aligning with industry needs. Many programmes already contain practical real-world problems as examples in their lecture material, but traditional capstone projects do not provide enough flexibility as student cohorts grow and diversify. The solution is to restructure projects around the concepts of work-integrated learning and project-based learning, providing distinct pathways tailored to different engagement modes: an industry-based internship, an industry-based research project or a university-based research project for problems that aren’t limited to one industry.

The industry-based internship focuses on practical business outcomes with immediate value to the industry partner (such as developing scalable data pipelines or predictive models), the industry-based research project emphasises longer-term innovations (for example, novel algorithms for asset management), and the university-based research project enables students to pursue research questions with more freedom and a potentially broader application (neural network architectures for large language models). This framework allows us to match project requirements with student capability, enabling better outcomes for the student, the industry partner and the university.

How success is measured is adapted accordingly. Academic assessments and rubrics are tailored to the learning outcomes, with some elements in common and others specific to each mode. All projects require students to submit weekly updates, reflecting on their progress and experience, and to deliver presentations and posters at the end of the summer. The industry-based internships require students to write an essay that combines technical and reflective writing. The research projects require students to write a white paper that balances academic rigour with practical insights. The industry-based projects use structured surveys to capture qualitative and quantitative feedback about the student and the project outcomes.

Structured industry engagement increases students’ workforce readiness, while supporting student autonomy fosters innovation and creates space for diversity, as research into work-integrated, project-based learning confirms.

Implementing a supervision model that scales effectively

The success of a well-designed curriculum relies heavily on how students are supervised. Traditional master-apprentice supervision collapses at scale and with diversity. It privileges students who already know how to perform and burns out supervisors. In practice, supervision at scale requires collaboration, co-learning and distributed expertise – with peer learning playing a central role, and the act of supervision itself understood as pedagogical, relational and socially situated.

Our tiered supervision model features central coordination for strategic oversight and consistency, primary supervisors for academic direction, secondary supervisors for hands-on support, dedicated pastoral care for student welfare and industry mentors for domain-specific insights. This structure distributes responsibility without diluting accountability, reducing reliance on a single supervisory relationship and making support more accessible for students who may be less familiar with academic norms or less confident asking for help.

Students are supported to move from guided practice to independent problem-solving, while supervisors can intervene early when risks emerge. Supervision becomes a shared teaching exercise rather than an individual burden, enabling the programme to scale without eroding quality.

Standardising processes for equity and efficiency

High-quality project-based learning requires standardised expectations across projects, not bespoke rules for each partner. Standardisation reduces cognitive load and reveals the “hidden curriculum”, which is important for students navigating unfamiliar cultural, academic or professional norms.

We use common scaffolds across all projects: scoping templates requiring early clarity on problem statements, stakeholders and success criteria; weekly updates capturing progress, blockers and reflections; and consistent governance around ethics, data handling and intellectual property. These represent what STEM practice looks like in the professional world. This approach aligns with work-integrated learning methodology. In practice, it means using case studies and reflective practice, with alignment between research questions, context and evidence.

Operational decisions also become teaching opportunities. Clear entry requirements ensure students and projects are matched appropriately. Streamlined legal workflows using templated agreements allow partnerships to form quickly. Plain-language project descriptions let students identify projects based on interest rather than insider knowledge. These choices reduce reliance on informal networks and widen access for students regardless of their culture or background.

Inspiring outreach and removing barriers to success in STEM

When these elements come together, the benefits compound. Students build professional confidence and networks supporting their transition into a career or further study. Industry partners gain innovation capacity with minimal risk, often leading to multi-year collaborations. The university strengthens its reputation as a grounded, informed research partner. Interdisciplinary work in fields such as health analytics, geospatial information systems and digital humanities also demonstrates to prospective students that STEM is not just a narrow specialism, but a comprehensive set of tools for understanding and improving the world.

Challenges remain – partners vary in readiness, projects can encounter scope changes or incomplete data, and students can be overwhelmed by complex or competing requirements. But standardised processes help navigate these realities, and working through ambiguity is itself valuable preparation for professional practice.

James Williams is a senior lecturer and data science industry coordinator at the University of Canterbury, New Zealand.

If you would like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.